This summer, with the support of the Dill Fund, I undertook a comprehensive research endeavor to delve into the complex interplay of transnational financial movements. Although my research heavily involved merging natural language processing (NLP) methods with econometric frameworks, its core was profoundly anchored in the real-life experiences of the Indian community in Canada. Brampton, Mississauga, and Toronto became my canvas. The bustling buses, the rhythmic hum of the subway, and even the local fast food chains were my classrooms. With every conversation, I was not just collecting data but piecing together a mosaic of experiences. Each conversation was an opportunity to delve deeper into the economic factors influencing their lives, earnings, and subsequent remittance levels. From these interactions, patterns emerged. The software engineer remitting funds for familial obligations, the student meticulously saving to offset an education loan, and the entrepreneur navigating the challenges of a new market, all provided invaluable insights. These narratives, rich and varied, were instrumental in refining the quantitative aspects of my research. The cultural nuances that I encountered during my fieldwork were enlightening. The Indian diaspora in Canada, while deeply rooted in their cultural heritage, has also embraced the Canadian ethos, creating a unique blend of
traditions and values. This cultural amalgamation has a profound impact on their financial behaviors and decisions. For instance, the significance of festivals like Diwali in influencing remittance spikes was a revelation. Such occasions not only represent cultural significance but also economic implications, as families back in India often rely on these remittances for festive expenditures. The initial SARIMA model I developed, while technically sound, benefited immensely from the sentiment scores derived from NLP analysis of news articles and my firsthand interactions. The dramatic improvement in forecasting accuracy, from a mean absolute percentage error of 62.96% to a mere 6.25%, underscored the significance of integrating qualitative factors into statistical models. This summer taught me more than just improved forecasting skills. It taught me about the resilience of the Indian diaspora, their dreams, and their challenges. The confluence of computational linguistics and econometrics, when juxtaposed with human narratives, opened up new vistas in financial forecasting, underscoring the value of a truly interdisciplinary approach.
